Skip to content
SimpleAct Logo

EU AI Act · Timeline

As of 3 October 2026 · after the Digital Omnibus

EU AI Act deadlines: every date after the Digital Omnibus

The AI Act has been in force since 1 August 2024, but its obligations apply step by step. The Digital Omnibus – Regulation (EU) 2026/1744, in force since 27 July 2026 – moved several dates: high-risk AI under Annex III now applies from 2 December 2027, product-integrated AI under Annex I from 2 August 2028. Here is every date with its legal basis, who it affects and what to do before it.

Reg. (EU) 2024/1689 · Reg. (EU) 2026/1744 · Sources linked on EUR-Lex

app.simpleact.de

Deadline monitor

EU AI Act · all dates

A

Next deadline

End of the marking transition, new prohibitions

2 December 2026

14days

  • End of the marking transition, new prohibitions2 December 2026 · 14 days
  • Legacy GPAI models and AI sandboxes2 August 2027 · 257 days
  • High-risk AI under Annex III2 December 2027 · 379 days
  • High-risk AI in regulated products (Annex I)2 August 2028 · 623 days

EU AI Act · Timeline

The timeline on one axis

Spacing follows real time. Filled dots already apply, hollow ones are ahead, the red one is the main deadline for high-risk AI.

  1. 1 August 2024In force

    AI Act enters into force

    Regulation (EU) 2024/1689 is in force. No obligations apply yet – the date starts the staggered transition periods.

    Art. 113Everyone

  2. 2 February 2025In force

    Prohibited practices and AI literacy

    The Art. 5 prohibitions apply, e.g. social scoring, manipulative techniques, emotion recognition at work and untargeted scraping of facial images. From the same day, providers and deployers must take measures on their staff’s AI literacy (Art. 4).

    Art. 4, Art. 5Providers and deployers

  3. 2 August 2025In force

    General-purpose AI models

    Obligations for GPAI model providers apply: technical documentation, information for downstream providers, a copyright policy and a training-data summary; for systemic-risk models additional evaluation and risk mitigation. The governance structure and the penalty framework apply as well.

    Ch. V, Ch. VII, Art. 99GPAI model providers

  4. 27 July 2026In forceOmnibus

    Digital Omnibus enters into force

    Regulation (EU) 2026/1744 moves the high-risk deadlines and rewrites Art. 4: companies must take measures to promote AI literacy but no longer guarantee a specific level per person. SMEs and small mid-caps get relief on documentation and quality management.

    Reg. (EU) 2026/1744Everyone

  5. 2 August 2026In force

    General application and transparency obligations

    The Regulation applies in general. The Art. 50 transparency obligations kick in: AI that interacts with people must say so, deepfakes and emotion recognition must be disclosed, and newly placed generative systems must mark their outputs in a machine-readable way. From now on the Commission can fine GPAI providers.

    Art. 50, Art. 101, Art. 113Providers and deployers

  6. 2 December 2026UpcomingOmnibus

    End of the marking transition, new prohibitions

    Generative AI systems already on the market before 2 August 2026 must now mark their outputs in a machine-readable way too (Art. 50(2)). Two new prohibitions apply: AI that creates or manipulates intimate depictions of identifiable people without consent, and AI that generates child sexual abuse material.

    Art. 50(2), Art. 5 (new)Generative AI providers, everyone

  7. 2 August 2027UpcomingOmnibus

    Legacy GPAI models and AI sandboxes

    GPAI models placed on the market before 2 August 2025 must meet the Chapter V obligations. Every Member State must run at least one national AI regulatory sandbox – the Omnibus pushed this back by a year.

    Art. 111(3), Art. 57GPAI providers, Member States

  8. 2 December 2027Main deadlineOmnibus

    High-risk AI under Annex III

    The high-risk requirements (Chapter III, Sections 1 to 3) apply to stand-alone systems under Annex III – e.g. in employment, education, creditworthiness, critical infrastructure and access to essential services. Providers need risk management, technical documentation, conformity assessment and registration; deployers must ensure human oversight and keep logs, among other duties (Art. 26).

    Art. 6(2), Annex III, Art. 8–27Providers and deployers of high-risk AI

  9. 2 August 2028UpcomingOmnibus

    High-risk AI in regulated products (Annex I)

    AI systems that are a safety component of a product under the Annex I harmonisation legislation, or are such a product themselves – e.g. machinery, medical devices, toys or lifts – must meet the high-risk requirements. Before the Omnibus this was August 2027.

    Art. 6(1), Annex IManufacturers and providers of regulated products

  10. 2 August 2030Upcoming

    Legacy high-risk AI at public authorities

    Providers and deployers of high-risk AI systems intended for use by public authorities must comply by now at the latest – even if the systems were in use before.

    Art. 111(2)Public authorities and their providers

EU AI Act · Timeline

Every deadline at a glance

The complete overview for reference – with legal basis and who is affected.

DateWhat appliesLegal basisAffectedStatus
1 August 2024AI Act enters into forceArt. 113EveryoneIn force
2 February 2025Prohibited practices and AI literacyArt. 4, Art. 5Providers and deployersIn force
2 August 2025General-purpose AI modelsCh. V, Ch. VII, Art. 99GPAI model providersIn force
27 July 2026Digital Omnibus enters into forceReg. (EU) 2026/1744EveryoneIn force
2 August 2026General application and transparency obligationsArt. 50, Art. 101, Art. 113Providers and deployersIn force
2 December 2026End of the marking transition, new prohibitionsArt. 50(2), Art. 5 (new)Generative AI providers, everyoneUpcoming
2 August 2027Legacy GPAI models and AI sandboxesArt. 111(3), Art. 57GPAI providers, Member StatesUpcoming
2 December 2027High-risk AI under Annex IIIArt. 6(2), Annex III, Art. 8–27Providers and deployers of high-risk AIMain deadline
2 August 2028High-risk AI in regulated products (Annex I)Art. 6(1), Annex IManufacturers and providers of regulated productsUpcoming
2 August 2030Legacy high-risk AI at public authoritiesArt. 111(2)Public authorities and their providersUpcoming

Digital Omnibus

What the Omnibus moved – and what it didn’t

Many overviews online still show the old dates. Regulation (EU) 2026/1744 was published in the Official Journal on 24 July 2026 and has applied since 27 July 2026. This is what changed:

High-risk AI under Annex III2 August 20262 December 2027
High-risk AI under Annex I (products)2 August 20272 August 2028
Machine-readable marking (Art. 50(2)) for existing systems2 August 20262 December 2026Newly placed systems: still from 2 August 2026
National AI sandbox per Member State2 August 20262 August 2027
AI literacy (Art. 4)Ensure a sufficient levelTake measures to promote itNew wording applies since 27 July 2026
Bans on non-consensual intimate deepfakes and abuse material–2 December 2026
Registration for the Art. 6(3) exemptionRegister in the EU databasestill required, simplified procedure

Unchanged: the Art. 5 prohibitions (since February 2025), the GPAI obligations (since August 2025) and the Art. 50 transparency obligations for new systems (since August 2026).

Deadline finder

Which deadline applies to your AI system?

Three answers are enough. The finder shows the dates that matter for your role and use case.

1. Your role
2. What applies to the system? · Select all that apply
app.simpleact.de

Your relevant dates

As of 3 October 2026 · after the Digital Omnibus

A

Choose your role to see the dates.

Guidance based on your answers, not legal advice. The binding classification depends on the specific system.

Already binding

What applies today – Omnibus or not

The postponement mainly concerns high-risk AI. These obligations are already running and are supervised:

  • Art. 5

    Prohibited practices

    Since 2 February 2025, with two new bans from 2 December 2026. The highest fine tier applies: up to €35M or 7% of worldwide annual turnover.

  • Art. 4

    AI literacy

    Since 2 February 2025, reworded since 27 July 2026. Companies must take measures – training, guidelines, a point of contact – and should document them.

  • Ch. V

    GPAI models

    Since 2 August 2025 for new models. Since 2 August 2026 the Commission can impose fines. Legacy models have until 2 August 2027.

  • Art. 50

    Transparency

    Since 2 August 2026. The marking transition for existing generative systems runs until 2 December 2026.

Backward planning

Planned backwards from 2 December 2027

About 13 months remain until the main deadline for high-risk AI. An example plan for a company with several AI systems:

  • Build the AI inventory

    Capture every AI system – including purchased tools and shadow AI.

  • Determine risk classes

    Annex III, Annex I, Art. 50 or minimal – with a rationale.

  • Roles and owners

    Provider or deployer? Who signs off per system?

  • Documentation and risk management

    Technical documentation, data quality, human oversight.

  • Conformity assessment or deployer duties

    Assessment, logging, fundamental rights impact assessment where required.

  • Registration and buffer

    EU database entry, rework, sign-off.

An example, not a requirement of the Regulation. Actual effort depends on the number and kind of your systems.

Special cases

Existing systems, authorities and SMEs

Systems already in use
High-risk AI placed on the market or put into service before the relevant date generally only falls under the new requirements if it is significantly modified afterwards (Art. 111(2)). A new version with a different purpose or logic can be such a change.
Public authorities
High-risk AI intended for use by public authorities has no permanent exemption: it must comply by 2 August 2030 at the latest.
Art. 6(3) exemption
If you classify an Annex III system as not high-risk because it only performs a narrow auxiliary task, you must document this and still register the system – via a simplified procedure in future.
SMEs and small mid-caps
The Omnibus eases technical documentation via a simplified Commission form and allows proportionate quality management. The deadlines themselves are the same for SMEs.
Large-scale EU IT systems
AI components of the large-scale IT systems listed in Annex X (e.g. in freedom, security and justice) placed on the market before 2 August 2027 must comply by 31 December 2030.

All dates in your calendar

Download the deadlines as a calendar file – for Outlook, Google Calendar or Apple Calendar. No sign-up.

Download calendar (.ics)

FAQ

Frequently asked questions about the deadlines

No. The Digital Omnibus moved the deadline for high-risk AI under Annex III to 2 December 2027, and for Annex I to 2 August 2028. On 2 August 2026, however, the Art. 50 transparency obligations and the general application of the Regulation started.
Yes. Regulation (EU) 2026/1744 was published in the Official Journal on 24 July 2026 and entered into force on 27 July 2026. The new dates are binding.
High-risk AI placed on the market or put into service before the relevant date generally only has to meet the new requirements if it is significantly modified afterwards. Exception: systems for public authorities must comply by 2 August 2030. For generative systems, the Art. 50(2) marking has a transition until 2 December 2026.
Yes, the dates are the same for everyone. SMEs and small mid-caps benefit from relief though: simplified technical documentation, proportionate quality management and lower fine ceilings.
The penalty framework has applied since 2 August 2025. Fines attach to the specific obligation: breaches of the prohibitions can be sanctioned since February 2025, breaches of high-risk obligations only once those apply. Ceilings under Art. 99: €35M or 7% for prohibited practices, €15M or 3% for other obligations, €7.5M or 1% for incorrect information – for SMEs the lower amount.
Yes. Instead of ensuring a sufficient level of AI literacy, providers and deployers must take measures to promote it since 27 July 2026. A specific level per person no longer has to be guaranteed. Training and its documentation remain the practical evidence.
Check two things: if you provide generative AI that was on the market before 2 August 2026, it must mark its outputs in a machine-readable way by 2 December 2026. And no system you use may fall under the new bans on non-consensual intimate deepfakes or abuse material.
Hardly, for high-risk AI. Inventory, classification, documentation and conformity assessment build on each other and usually take several months. Inventory and risk classification pay off early anyway, as Art. 4, Art. 5 and Art. 50 already apply.

More questions? We're happy to help. Send email · Try 7 days (credit card required)

Sources and status

Last reviewed on · SimpleAct editorial team

This page is for orientation and is not legal advice. The wording in the Official Journal of the EU is authoritative.

Changes to this page

  • 10 October 2026 – Entry into force (1 Aug 2024) and staged application clarified.
  • 3 October 2026 – Rebuilt: Omnibus entry into force, Art. 50(2) transition, new prohibitions, dates for sandboxes, legacy GPAI models and public authorities added.
  • 2 October 2026 – High-risk deadlines updated after the Digital Omnibus.

Deadlines in view, evidence in the system

SimpleAct maps every AI system to its risk class and the dates that apply – with owners, documentation and audit history.